In Sections 6.3.3 and 6.3.2 it was indicated that a relationship between the volumetric water content of the upper soil layer and the topography exists.
In this section this relationship is further analyzed andKa within Ladegaards Enge 3 June 1997 is reconstructed. In addition a possible connection between the volumetric water content of the upper soil layer in Ladegaards Enge and the discharge in the river during periods of low precipitation is investigated.
6.4 Fusion of topography and Ka 143
Figure 6.19: Variations inKa, the volumetric water content and the level along transect T1 at Ladegaards Enge. Kais based on TDR measurements from June 1997 and July 1999. The estimated volumetric water content is derived from soil samples from 1999 and based on Ka from 1999 derived from Topp et al (1980). The level corresponds to the height above sea level.
6.4.1 Ladegaards Enge 1997
In Figure 6.19 a relation between the topography and the volumetric water content θw along transect T1 is evident. Here θw is derived from soil samples gathered 12 July 1999 and listed in Table6.3.
In order to test the strength of the monotonic association between θw and the topography we apply Spearman’s rank order correlation test. This is a dis-tribution free test that determines whether or not a monotonic relation exists between two variables. Because linear relations very often will be unrealistic in practical situations Spearman’s test is more appropriate in this context than Pearson’s correlation test.
Spearman’s rank order correlation coefficient rs is given by
rs= 1− 6D
n3−n, (6.1)
Topp sandy loam + fibric/hemic peat
Ka
Figure 6.20: Correlation between the estimates of the volumetric water content θw in the soil samples and the estimates of the corresponding Ka values from 12 July 1999 in Ladegaards Enge. The samples are collected along transect T1.
The solid black curve is a least squares fit using a third-order polynomial. The solid grey curve shows the relation by Toppet al. (1980).
whereD is the sum of the squared differences in rank order constants andnis number of observations. The distribution ofDis symmetrical around (n3−n)/6, it is only defined for even numbers and 0≤D≤13(n3−n). The distribution is given under the constraint that no correlation exists, which is ourH0hypothesis.
Using (6.1) the correlation coefficientrsbetweenθwand the topography is−0.77 for n= 10 andD = 292. We therefore have 1−P(D ≥292) = 0.0063 which implies that H0 is rejected using the significance level 0.05. In other words we may conclude thatθw and the topography are correlated.
In Figure6.19there also is an almost linear relationship between the apparent di-electric constantKa(1999) andθwderived from the soil samples. In Figure6.20 these values are plotted against each other and the third-order polynomial re-lationship is estimated using a least squares fit. This plot again suggests a relationship and it would therefore be relevant to test to what extentKa(1999) values andθware correlated.
6.4 Fusion of topography and Ka 145
Figure 6.21: Scatter-plot of the kriged topography in Figure 6.15 against the estimatedKaderived from the TDR-measurements performed 3 June 1997 and 12 July 1999 at Ladegaards Enge. The height is metres above sea level and the solid lines are the best fitted third-order polynomials in a least squares sense.
We again use (6.1) to estimate the correlation coefficient and get rs = 0.64 D= 60 andn= 10. This results inP(D≤60) = 0.0272 which means that it is unlikely that no correlation exists. Based on that we conclude thatθw andKa
are associated.
Given the correlations and test statistics above strong evidence suggests that the topography and Ka at T1 3 June 1997 are correlated. It therefore seems relevant to test the correlation between the topography and Ka values derived from TDR-measurements performed 3 June 1997. This is possible due to the kriged topography in Figure 6.15, which enables estimates of the topography at every location where TDR-measurements are performed. The constraint that the topography has not changed in the period from 3 June 1997 until the measurements of the topography in the fall of 1998 is fulfilled.
In Figure 6.21 the kriged topography at Ladegaards Enge is plotted against the Ka estimates from 3 June 1997. Applying Spearman’s rank order corre-lation test (6.1) we get D = 165670, n = 81 and rs = −0.87. Because n is high,Dis approximately Gaussian distributed and the test statistic is therefore
20 40 60 80 100 120
Figure 6.22: A map showing the krigedKa in the test site at Ladegaards Enge 3 June 1997. The resolution is 0.25 m.
1−Φ(7.8)<0.00003. This rejects theH0 hypothesis and we conclude thatKa
is correlated with the topography.
In order to establish a relation between the topography and the Ka estimates a third-order polynomial in a least squares sense is fitted to the points in the scatter-plot in Figure6.21. The polynomial relationship is
Ka= 274305.04−41581.42h+ 2099.20h2−35.289h3, (6.2)
where h is the height above sea level. The polynomial relationship between the topography and Ka is only valid for the test area at Ladegaards Enge 3 June 1997. Based on (6.2) and the kriged topography in Figure 6.15 a 2-D representation of the variation ofKa 3 June 1997 is constructed and illustrated in Figure6.22.
6.4 Fusion of topography and Ka 147
Figure 6.23: Discharge and temperature in Ladegaards Enge from 1 June 1999 to 30 July 1999. The precipitation in the period is negligible. The additional fieldwork was performed 12 July 1999. The river discharge is measured at the gauging station in Ladegaards Enge and the temperature and precipitation are averaged over 24 hours within an area of 20 km × 20 km. (Copyright Danish Institute of Agricultural Science. The hydrometric data are from NERI’s gauging station at Sminge Vad).
6.4.2 Ladegaards Enge 1999
In Section 6.4.1 it was shown that the topography and Ka were correlated 3 June 1997 at Ladegaards Enge. It therefore would be interesting to explore to what extent Ka in general is governed by the topography and to what extent Ka is affected by the discharge in the river.
In this section we therefore will test the strength of the monotonic associa-tion between the topography and the Ka estimates derived from the TDR-measurements performed 12 July 1999. Figure 6.21illustrates the plot of the topography againstKa and the solid curve represents a least squares fit using a third-order polynomial. The fitted third-order polynomial in the least squares
sense is
Ka= 487163.93−75812.97h+ 3932.97h2−68.01h3, (6.3)
wherehis the height above sea level. This relationship is valid for the test area in Ladegaards Enge 12 July 1999.
Spearman’s rank order correlation test (6.1) is again applied and the calculated statistics are n = 175,D = 1574235 and rs =−0.76. Thereby H0 is rejected andKa is again correlated with the topography.
When comparing the two plots in Figure6.21it is obvious thatKa, and thereby θw, is lower 3 June 1997 than 12 July 1999 within the upper soil layer. This is also in perfect accordance with the Figures 6.5 and6.23that state that the discharge in the river 3 June 1997 was 0.6273 m3/s and 12 July 1999 0.77 m3/s.
The assertion that the volumetric water content during periods of low precip-itation is related to the discharge in the river and the topography is thereby supported.
6.5 Discussion
The test site in Ladegaards Enge is a part of the Gjern catchment in the Eastern Jutland. It is a riparian wetland where the water table is high and approximately intersecting the ground surface at the river bed.
Within the area the various plant communities are strongly correlated with the soil moisture content. In the drier part of the test site corresponding to sub-area III Alopecurus pratensisis prevailing and in the more humid sub-area II Deschampsia caespitosa is widely distributed. Ellenberg (1992) has classified the plant species on a scale from 1–12 in accordance to their preference for humidity [28]. Here Alopecurus pratensisscores 6 andDeschampsia caespitosa 7, which supports our experiences from the field that soil moisture and plant communities are related.
The deposits in the upper 8 cm in the floodplain is fibric to hemic peat, which is overlaying a cleyic loam. The coarser fraction of sandy loam is deposited at or near the levee. These deposits originate from overflows and the meandering of the river. It is therefore natural that the organic content and fine-textured soil particles increase away from the river as Table6.3indicates.
6.5 Discussion 149
Fine-textured soil particles such as clay and organic matter bind the water mole-cules, which results in a lowerKa than if the soil had been coarse textured [39].
According to Table 6.3 the organic content increases along T1 away from the river and as a consequence one might expect the estimated Ka within Lade-gaards Enge to be affected by the variability of the organic content. However, according to Figure 6.19the increase in the organic content along T1 does not seem to affect the estimatedKa. This is because the microwave frequencies are strongly sensitive to water compared to the soil particles. Here theKavalue for water is approximately 80 whereas it is only 2–4 for the other soil constituents [56]. Because the general conditions were quite humid 12 July 1999, especially in the area where the organic content was high, the contribution from the fine textured soil and organic content was negligible. The soil moisture conditions were also high 3 June 1997 and it is therefore assumed that the derivedKafrom the fieldwork 3 June 1997 is unaffected by the organic content within Ladegaards Enge.
In Section 6.4.1 and 6.4.2 it was shown that Ka is strongly correlated with the topography within Ladegaards Enge. This was in Section 6.4.1utilized in the making of the map in Figure 6.22, which covers Ka 3 June 1997. When examining the plots in Figure 6.21it is obvious that the spatial distribution of Ka 3 June 1997 and 12 July 1999 within the test site is different. The general level of Ka at the test site 12 July 1999 is higher than 3 June 1997 and the variability of Ka 12 July 1999 less than 3 June 1997. This suggests that the best fitted line in the figures approaches the level of 80, which is theKa value for water at 20◦C, as the general soil moisture increases within the test site [35].
It is therefore necessary to have some prior knowledge of Ka in order to utilize the topography for estimatingKa within the test site at a given time.
Chapter 7
Mols Bjerge
Fieldwork was performed at Mols Bjerge 4 June 1997 at three test sites in order to enable the study of the synergy between semi-natural grasslands and the restored polarimetric EMISAR data in Chapter9.
The test areas of investigation are located at Trehøje, Benlighøj and Stenhøje and the in situ data collected comprise an evaluation of the vegetation char-acteristics, estimation of biomass, soil samples and TDR-measurements. These in situdata represent factors that are all known to affect the polarizations and frequencies used by EMISAR. For a brief description of the sampling method-ologies and the interaction between the in situ data and EMISAR refer to the Sections3.1.1, 3.1.2 and 3.1.3. In this chapter the sampling strategy at each test site is outlined and thein situdata are presented and analyzed.
A lot of research has earlier been performed on grasslands at Mols Bjerge. For a study of the spectral identification of plant communities for mapping of semi-natural grasslands refer to Jacobsen (2000) [38].
Figure 7.1: The geographical placement of Mols Bjerge. The blue arrow shows the flightline of the EMISAR where the acquisitions are made within the start and end points on 3 and 4 June 1997. The EMISAR is looking to the left and the red spot to the South indicates the test sites Trehøje and Benlighøj. The red spot to the North indicates Stenhøje. (Map material from the Danish Kort-og Matrikelstyrelsen (KMS) is reproduced according to agreement G18/1997 between NERI and KMS).
7.1 Description of test sites
Mols Bjerge is located at the peninsula in the southern part of Djursland in Jutland. It is a beautiful area where the hills are covered with heath, grassland and forest. Geologically Mols Bjerge is a terminal glacial moraine from the last glaciation Weichsel (W¨urm). The moraine is a mixture of sand, clay and pebbles [64], [67].
Figure7.1shows the geographical placement of Mols Bjerge as well as the flight-line of the EMISAR and the three test sites. The test areas of study are located at Trehøje and Benlighøj, which are indicated by the southern red spot, and Stenhøje indicated by the northern red spot.
In Figure7.2two of the test sites are indicated on an ortho-photo from 1995.
7.1 Description of test sites 153
Figure 7.2: An aerial view of two of the test sites at Mols Bjerge displayed on an ortho-photo from 1995. The test area Trehøje is located within the four red crosses to the left. The four red crosses to the right represent the test site Benlighøj. (Ortho-photos are copyright Kampsax 1995).
The four red crosses to the left comprise the Trehøje test area whereas the four red crosses to the right represent the test site Benlighøj. Likewise, the displayed crosses in Figure 7.3 represent the test site at Stenhøje. The ortho-photos are from 1995 and originally geometrically rectified according to system 34 for Jutland. However for practical purposes, the coordinate system used in the following is Universal Transverse Mercator (UTM), zone 32, datum ED50.
The test areas are all classified as grasslands by Jacobsen (2000) [38]. The Trehøje test site illustrated in the photo in Figure7.4 (a) is an old abandoned grassland, which was dominated by Deschampsia flexuosa. Stenhøje test site was characterized by the green and vigorous vegetation of Festuca rubra as displayed in the photo in Figure7.4(b). The photos in Figure7.5show the test area at Benlighøj, which was grazed and dominated byDeschampsia flexuosa.
The criteria for selecting the test sites for this study have been homogeneity in terms of soil moisture, above ground biomass and vegetation characteristics within each site. Between the sites differences exist in the plant species, the volumetric structure of vegetation and most likely in the biomass content. Al-though Mols Bjerge is a very hilly area the three selected test sites are flat and
Figure 7.3: An ortho-photo illustrating the test site Stenhøje at Mols Bjerge 1995. The test area is located within the red crosses. (Ortho-photos are copy-right Kampsax 1995.)
orientated towards the EMISAR in such a way that the incidence anglesϕfor the EMISAR images within Trehøje, Benlighøj and Stenhøje are almost identi-cal. At the test areas Trehøje and Benlighøj the localϕis 39◦ whereas the local ϕat Stenhøje is 41◦.
At the time the fieldwork was performed the weather was calm, sunny and hot and dry conditions had been prevailing for some weeks. In Figure7.6 is shown the temperature and precipitation at Mols Bjerge from 1 May 1997 to 30 June 1997. Although the figure gives a hint of the weather conditions prior to the fieldwork on 4 June 1997 at Mols Bjerge it should be noted that the temperature and precipitation are averaged over 24 hours within an area of 40×40 km.